BOX 2-2

The Variability Hypothesis

Mean differences between men and women in scores on mathematics and science achievement tests are not especially large, and mean scores have been converging. Many believe that these trends are largely irrelevant, however, because people who go on to research careers in science, mathematics, and engineering are not drawn from areas near the midpoint of science and mathematics abilities, or the fat part of the bell curve. Instead, the assumption is often made that those who end up in research careers in science, engineering, and mathematics (SEM) are drawn from the top 1-5% of the distribution in mathematics and science talent.a

It is precisely at this extreme tail of science and mathematics abilities that sex differences are most evident. For example, in a study of close to 10,000 talented 12- and 14-year-olds who had taken the SAT, the male:female ratio was 2:1 for those with SAT-M scores of at least 500; it was about 12:1 for those with scores of at least 700.b Such findings are often viewed as part of a pattern of greater variability in ability and achievement among men than among women. As Steven Pinker has so succinctly stated, when it comes to male abilities and achievement there are “more prodigies, more idiots.”c

The variability hypothesis has a great deal of face validity and appeal. College-educated SEM professionals make up only 2-3% of the US workforce, so shouldn’t they be those in the top 2-3% in science and mathematics abilities? Interestingly, the answer to that question, often assumed, has not been examined until recently. And the answer appears to be no. A recent economic analysis by Weinberger examined characteristics of the college-educated SEM workforce and found that fewer than one-third of the white males had SAT-M scores above 650, which is at the low end of the threshold for ability in mathematics typically presumed to be required for success in these fields.d In both samples of adolescents followed in the analysis, about one-fourth of the college-educated men and women in the SEM workforce had SAT-M scores below the 75th percentile, and more than half the men (and almost half the women) had scores below the 85th percentile— much closer to the fat part of the curve than anyone had imagined.

Those findings cast serious doubt on the variability hypothesis as the cause for the large discrepancy between the numbers of men and women who go on to SEM careers. It should be noted that the Weinberger study included SEM workforce participants holding bachelors degrees and above, and did not address the subset of those who obtain SEM doctorates.

A further argument against the variability hypothesis stems from its malleabil

The National Academies of Sciences, Engineering, and Medicine
500 Fifth St. N.W. | Washington, D.C. 20001

Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement